Glycine Substitutions in Collagen Heterotrimers Alter Triple Helical

Jan 18, 2017 - Katherine A. Clements†, Amanda M. Acevedo-Jake†, Douglas R. Walker, and Jeffrey D. Hartgerink. Department of Chemistry, Rice Univer...
0 downloads 0 Views 7MB Size
Article pubs.acs.org/Biomac

Glycine Substitutions in Collagen Heterotrimers Alter Triple Helical Assembly Katherine A. Clements,† Amanda M. Acevedo-Jake,† Douglas R. Walker, and Jeffrey D. Hartgerink* Department of Chemistry, Rice University, 6100 Main Street, Houston, Texas 77005, United States S Supporting Information *

ABSTRACT: Osteogenesis imperfecta typically results from missense mutations in the collagen genome where the required glycine residues are replaced with another amino acid. Many models have attempted to replicate the structure of mutated collagen on the triple helix level. However, composition and register control of the triple helix is complicated and requires extreme precision, especially when these destabilizing mutations are present. Here we present mutations to a compositionand register-controlled AAB helix where one of the requisite glycines in the A chain of the triple helix is changed to serine or alanine. We see a loss of compositional control when the A chain is mutated, resulting in an A′BB composition that minimizes the number of mutations included in the triple helix. However, when both A and B chains are mutated and no nonmutated peptide chains are available, the designed A′A′B′ composition is reestablished. Our work shows the ability of the mutations to influence and alter the composition and register of the collagen triple helix.



INTRODUCTION A collagen triple helix is a supramolecular complex that consists of three peptide chains in the polyproline II conformation1,2 that self-assemble to create a right-handed superhelix. Each chain has a three-amino-acid repeat of Xaa-Yaa-Gly with stabilizing hydrogen bonds that form between a glycine NH and the backbone carbonyl of an Xaa position in an adjacent chain. In a triple helix the separate peptide strands may be the same or different sequences, which determines the composition of the triple helix. When all three peptides are identical the helix is called an AAA homotrimer, when there are two unique peptides the helix is an AAB heterotrimer, and when all three are distinct the helix is called an ABC heterotrimer. Because there is a one amino acid offset between each chain of the triple helix, the order of the peptides can create unique triple helical registers in heterotrimers. For example, an AAB triple helix will be distinct from an ABA because different chains will be in the leading, middle, and lagging positions. The design of composition and register of specific collagen triple helices has advanced, allowing for more complex models of disease conditions, specifically the mutations found in osteogenesis imperfecta (OI).3 OI is a disease that usually results from mutations of glycines in the Xaa-Yaa-Gly repeat in type I collagen. These mutations cause a wide range of disease phenotypes based on the identity of the amino acid replacing glycine and the location of the mutation in the sequence.4,5 From clinical data, experiments, and theoretical models, the severity of the disease has been correlated with the size and charge of the substituted amino acid, specifically Ala < Ser < Cys < Arg < Glu < Asp < Val, in order from least to most disruptive.4,6−8 The placement of the mutation in the A chain © XXXX American Chemical Society

or B chain also changes the severity because, in principle, two mutations are incorporated when the A chain is mutated, and only one is incorporated when the B chain is mutated.4,5 Before register and composition control were available, the first OI models were homotrimers. Although these systems gave significant insight into glycine mutations in collagen, they cannot accurately account for structural and thermal disruptions found in natural heterotrimeric mutations. For a glycine-to-alanine substitution in a homotrimer the mutation is always incorporated into all three chains of the triple helix.9 Among other changes to the helix structure, this has been shown to result in the loss of four intermolecular backbone hydrogen bonds. This local disruption decreases the melting temperature, Tm, of the helix by 29 °C compared with the base (POG)10 sequence. Further mutations have been incorporated into a homotrimeric triple helix and studied with circular dichroism and NMR.9−11 In homotrimers, generally, a mutation typically results in the loss of three to four backbone hydrogen bonds and a sharp decrease in thermal stability. With the development of ABC heterotrimeric triple helices with compositional control, mutations could selectively be made in just one or two chains. Previously, our lab prepared an ABC triple helix with a central AAB guest region to model mutations in one, two, and all three chains.12 We saw that once the triple helix was mutated, successive mutations had a smaller impact on the stability of the helix; specifically, the destabilization was not linear with respect to the number of mutations. However, Received: December 5, 2016 Revised: January 2, 2017

A

DOI: 10.1021/acs.biomac.6b01808 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Coupling cycles employed a 1:4:4:6 ratio of resin/amino acid/HATU/ diisopropylethylamine (DiEA). For coupling 15N-glycine isotopic labels, this ratio was reduced to 1:2:2:3. Acetylation at the N-terminus was performed twice with a mixture of resin/DiEA/acetic anhydride (1:10:5) in dichloromethane. A standard cleavage cocktail containing EDT/H2O/TIPS (1:1:0.4) in 20 mL of trifluoroacetic acid (TFA) was performed for 3 h to remove the peptides from the resin. The TFA was removed using rotary evaporation; then, the product was precipitated out with cold diethyl ether. The precipitate was pelleted using centrifugation, and the ether wash was repeated two additional times. Peptide Purification. The peptides were purified with a combination of dialysis and HPLC. Dialysis was performed using SpectraPor tubing with a 500−1000 MW cutoff for 3 days. Then, the solution was diluted to roughly 5 mg of peptide/1 mL of H2O, and purified through reverse-phase high-pressure liquid chromatography (HPLC). HPLC was performed on a Varian PrepStar220 with a Phenomenex Jupiter 4 μ Proteo 90 Å column. A gradient of 1% acetonitrile (with 0.05% TFA) in water (with 0.05% TFA) was used, and the main peak was confirmed using electrospray ionization timeof-flight mass spectrometry (ESI TOF-MS). MS and HPLC spectra are available in Figures S1−S6. The TFA and acetonitrile were removed via rotary evaporation, and the sample was frozen and lyophilized to give pure peptide powder. Sample Preparation. Stock solutions of peptides were prepared with Milli-Q water, and the pH was adjusted to 7 with sodium hydroxide. For circular dichroism the samples were diluted to 0.3 mM peptide with 10 mM phosphate, and for NMR, the samples were prepared at 3 mM total peptide concentration and 10 mM phosphate. These solutions were annealed at 85 °C and allowed to incubate for 1 week at room temperature and 1 week at 4 °C. Circular Dichroism Spectroscopy. A Jasco J-810 spectropolarimeter equipped with a Peltier temperature controller was used to record all secondary structure spectra and thermal unfolding studies. Spectra were collected between 190 and 250 nm. Melting experiments monitored the maxima around 225 nm from 5−85 °C at a heating rate of 10 °C/h, and the minimum of the first derivative of the melting curve was used to determine the melting temperature of the sample. The molar residual ellipticity (MRE), shown as [θ], was calculated from the recorded ellipticity using

this system gave no information about composition or register and also lacked structural characterization of the mutation site. Recently, utilizing a robust design of an AAB register and composition controlled triple helix, just the B chain of the triple helix was mutated to study the impact on the triple helix stability and structure.13 This system showed a clear correlation between the structural disruption and the size or charge of the substituted amino acid. Smaller uncharged amino acids exhibited less disruption to the triple helix than large charged amino acids. The mutations also showed an impact on the composition and register, where increasing size or charge of the mutated amino acid led to disruptions away from the desired AAB′ population. Here we complement that study by analyzing the impact of mutations to the A chain of AAB heterotrimers. We also examine systems in which both the A and B chains have been mutated. Our analysis shows the strong driving force for triple helices to avoid double mutations and the changes in helical composition that arise to accomplish this despite the loss of many stabilizing charge−pair interactions. Peptide Design. To mimic natural OI disruptions, we used a highly stable AAB register controlled system as our base sequence and introduced glycine mutations.14 To explore the molecular structure of the triple helices, 15N isotopically labeled glycines were placed around the mutated region and also were placed N-terminally (Figure 1) to give information on

[θ ] = θ /(C × 1 × nr × 10)

In eq 1 θ represents the ellipticity c is the sample concentration (M), l is the path length (cm), and nr is the number of residues in the peptide. The first derivatives of the melting curves are available in Figure S11, and the melting points are given in Table 1. NMR Spectroscopy. All TOCSY and 1H,15N-HSQC and 1H,15NNOESY-HSQC experiments were performed on a Varian spectrometer of 800 MHz containing a cryogenic probe. The nitrogen carrier frequency was set at 110 ppm, while the proton carrier frequency set to the water signal. All experiments were performed at 10 °C. Chemical shift assignments of residues were made using TOCSY spectra. For TOSCY spectra, 1120 × 512 complex points and 128 scans were acquired; the mixing time was 100 ms and sweep width was 12 kHz in both dimensions. The 1H,15N-HSQC were collected with 1040 × 256 complex points for a total of 16 scans. A sweep width of 10 kHz in the proton and 2432 Hz in the nitrogen dimension was used. To compare relative peak volumes of folded and unfolded states, for each 1H,15N-HSQC of an AAB heterotrimer the total peak volume was set to 1. Because of the difference in the relaxation rate of the monomer and folded helix, no quantitative relationship between the populations can be determined from this ratio, but a general comparison between the mutation types may be made (Table 1). A 2D version of the 3D 1H,15N-NOESY-HSQC was collected with 100 ms mixing time, 64 scans, 1200 × 256 complex points; the spectral window was 12 kHz in the direct dimension and 12 kHz for the indirect dimension. 15N relaxation experiments were performed on 3 mM samples, as previously described on a 500 MHz Varian spectrometer.13 Data were processed and analyzed using NMRdraw15 and ccpnmr16 processing suites. Additional spectra including 1H,15N-

Figure 1. Sequences, compositions, and registers of peptides used. Red amino acids are negatively charged, and blue are positively charged. The yellow boxed glycines are 15N isotopically labeled. A is the B-Ala mutant, B is A-Ala mutant which is observed to form, C is A-Ala mutant which was intended but is not observed to fold, and D is the 3Ala mutant. The serine mutants are identical other than the substitution of alanine for serine.

refolding. Additionally two different amino acid substitutions (alanine or serine) were tested. Here systems with a mutation only in the A chain are called A-Ala or A-Ser depending on the amino acid substituted. The double-mutant systems are called 3-Ala and 3-Ser.



(1)

EXPERIMENTAL SECTION

Peptide Synthesis. Peptide synthesis was performed on an Advanced Chemtech Apex 396 peptide synthesizer utilizing standard solid-phase peptide synthesis techniques with FMOC protecting groups. The peptides were synthesized on a 0.15 mM scale on Rink amide MBHA resin (0.37 mmol/gram loading). Deprotection cycles were carried out with 25% piperidine in dimethylformamide solution. B

DOI: 10.1021/acs.biomac.6b01808 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules Table 1. NMR Experimental Results of the Different Osteogenesis Imperfecta Model Systemsa substitution

name

heterotrimer Tm (°C)

folding

R2/R1 of heterotrimer

monomer rel vol.

AAB′ rel. vol.

glycine14 alanine13 serine13 alanine serine alanine serine

base B-Ala B-Ser A-Ala A-Ser 3-Ala 3-Ser

51.0 30.5 29.5 34.0 30.0 19.0 18.0

single composition, single register AAB single composition, single register AAB′ AAB′ folded as well as AAA homotrimer single composition, single register A′BB single composition, single register A′BB single composition A′A′B′ single composition, single register A′A′B′

14.9, 9.6, 21.3, 8.0 20.8, 10.7, 13.8, 10.0 15.7, 13.5, 18.6, 17.8 7.7, 8.9, 11.9, 7.8, 6.0 8.6, 11.1, 6.7, 10.4, 9.4 5.3, 2.4, 3.0, 4.5 8.3, 4.9, 5.3, 5.8

0.33 0.44 0.61 0.50 0.82 0.81 0.83

0.67 0.56 0.37

A′BB rel. vol.

0.50 0.18 0.19 0.17

a

Heterotrimer Tm is the melting point of the folded heterotrimer by CD. The R2/R1 value, or relaxation value, is for each isotopic label (G1−G5) in the order from leading to lagging chain. Relative volumes are the relative volume of the peaks when the total peak volume is set to 1 (see Experimental Section for more details).

Figure 2. 1H,15N-HSQC of the peptide systems. Black spectra are the AAB mixture. Blue is only the B strand used in the system. Red is only the A strand used in the system. G1−G5 denote the heterotrimer present and are labeled in the same order as shown in Figure 1. A1−A3 denote the AAA homotrimers. M1−M3 denote the peptide monomers. B-Ala and B-Ser spectra are reproduced from ref 13 for comparison. NOESY-HSQCs and TOCSY, along with a discussion of the assignment process, are available in the Supporting Information Figures S13−S23. Molecular Modeling. Models were created by first preparing a basic structure with the appropriate sequence and expected composition and register. For the B-Ala, B-Ser, 3-Ala, and 3-Ser helices, an AAB composition and register was used. For A-Ala and ASer, an ABB composition and register was used. Constraints were created for the heterotrimers from their assigned 1H,15N-NOESYHSQC spectrum. These constraints were used to model the heterotrimers using the Rosetta Relax application.17,18 Negative constraints were created for situations where constraints were present in the base structure but not the mutated structure. Negative constraints were not used for the three chain mutants because their initial structure deviated significantly from the base structure. Positive correlations were set for 0−4 Å and negative from 6 to 10 Å. Once a set of structures was created, they were analyzed for compliance to the constraints. Positive correlations were considered satisfied if they were below 5 Å, and negative correlations were considered satisfied if they

were above 5 Å. The most compliant structure was then iterated until all constraints were satisfied. If any constraints were still unsatisfied after multiple iterations, then the standard deviation of that constraint was decreased to increase the penalty for longer distances. The final structures presented here are compliant to all NMR constraints. The constraints used are available in Tables S1−S4. Images of the final models overlaid with the base structure are also available in Figures S7−S10. Helical Twist. A program was written in Java with a J/Link to Mathematica19 to perform the twist measure calculations in this study. For each residue the coordinates of the α carbon were considered for the purpose of conducting calculations. The program was used to extract α-carbon coordinates from two adjacent triads on the helical axis and perform a series of transformation calculations on these coordinates. Kabsch twist values,20 calculated in a similar manner as previously presented,21 were obtained by first calculating the covariance matrix for each pair of triads, then performing a singular value decomposition (SVD) of the covariance matrix in Mathematica and extracting the optimal rotation matrix according to the Kabsch C

DOI: 10.1021/acs.biomac.6b01808 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules algorithm. Values for translation and rotation were extracted from this matrix. Radius and displacement values were obtained from the difference of the α-carbon coordinates and the centroid of the triad. Distances between the triads were obtained by taking the difference of the coordinates of two adjacent centroids. The tilt angle was obtained by finding the angle between the perpendicular vectors of each triad. The direct twist values, as reported in this paper, were obtained by overlaying the centroids of the two adjacent triads, tilting the second triad by the tilt angle found above, and calculating the angle between a particular residue, the centroid, and the corresponding residue from the adjacent triad for each position. These methods are described more in-depth in the Supporting Information.

the base system. However, the A-Ala A′BB helix that folds has only five highly stabilizing charge pairs, while the B-Ala AAB′ helix has seven that form (Table 2). The increase in melting Table 2. Results of the Modeling Experimentsa



RESULTS AND DISCUSSION Base AAB Triple Helix. To improve upon the physiological relevance of OI models and gain better insight into natural systems, a previously designed register and compositioncontrolled AAB heterotrimer was used to mimic the composition of type I collagen.14 To ease composition verification in our systems, a single 15N-glycine label is included in the A chain and two 15N-glycine labels are included in the B chain. This produces four folded 1H,15N-HSQC peaks in an AAB triple helix as well as a peaks for both the cis and trans conformations of the labels in unfolded single A or B chains (Figure S12). This is a well-folded system, shown by the high ratio of trimers to monomers, and the relatively high (>8) relaxation values of the 1H,15N-HSQC peaks indicate good local stability. R1/R2 relaxation values below two typically indicate monomeric peptides, while higher values indicate increased local rigidity of a folded helix. The CD melting spectra corroborate the robustness of the base heterotrimer, showing a single transition at 51 °C. The helical twist of the base triple helix has a typical amount of variation and falls mostly within the range of a typical POG triple helix (Figure S12). It does have one point outside of the range, which is believed to be the result of strain induced by the surrounding charge pairs. A-Chain Mutants. Like the base triple helix, the triple helices A-Ala and A-Ser are mixed at a 2:1 ratio of A′:B and are designed to fold into an A′A′B triple helix (A′ indicates the mutated version of the A peptide) composition and register. As seen for the base triple helix, this composition should yield four peaks in the trimer region of the 1H,15N-HSQC spectrum corresponding to the four different isotopic labels in the trimer (Figure 1). Instead, however, the NMR showed five distinct trimer signals of equal volume for both the A-Ala and A-Ser cases, labeled G1−G5 in Figure 2. In our labeling scheme, these five peaks can appear only if the composition has been altered from the intended A′A′B composition to an A′BB composition. One peak represents the label in the A′ chain, and the B chains each contain two 15N labels. (Figure 2). While type I collagen is quite long (∼1000 residues), our shorter 30-mers are not able to compensate for the local disruption that would be induced by two glycine mutations in the A chain and instead preferentially form the A′BB composition. Our results suggest that the energetic barrier needed to compensate for two A chain mutants outweighs the stabilization that the A′A′B system could produce by folding; in this manner, natural systems may also be affected and suffer from poor folding equilibria. A-Ala melted at 34 °C, making it much less stable than the base Tm of 51 °C but also more stable than a previously reported B-Ala triple helix, which melted at 30.5 °C (Table 1, Figure S11). Both A-Ala and B-Ala have disrupted equilibria, as seen by the higher ratio of monomer to heterotrimer than in

triple helix

lost hydrogen bonds

lost charge pairs

compensating interactions

formed charge pairs

B-Ala13 B-Ser13 A-Ala A-Ser 3-Ala 3-Ser

5 2 1 1 1 2

3 4 0 0 1 1

8 6 0 5 3 7

7 6 5 5 9 9

a

Lost hydrogen bonds are backbone hydrogen bonds that are expected to be found in the model but are not present. Lost charge pairs are designed axial charge pairs that are not formed in the model. Compensating interactions are hydrogen bonds that are not present in the base structure but are formed in the model. Formed charge pairs are designed axial charge pairs that are present in the model.

point is in contrast with the relaxation values (R2/R1) for the isotopic labels (7.7, 8.9, 11.9, 7.8, and 6.0) (relaxation values are in order from leading to lagging chain), which are also lower than both the base system and B-Ala, indicating that A-Ala is a less ordered helix and is likely to have a more disordered conformation around the mutation (Table 1). Commonly, relaxation values are used to measure the quality of triple helical folding, and it is intuitive to assume that a well-folded helix would have a higher melting point than one with lower relaxation values. However, we see the opposite, where our less well-folded helix A-Ala has a higher melting point than our better-folded helix, B-Ala. This suggests that the melting point is not comprehensive in accessing the quality of a triple helix, and other methods such as NMR and molecular modeling must be used. A model was created of the A-Ala A′BB heterotrimer utilizing constraints from NMR. When compared with the B-Ala AAB′ model, the A-Ala A′BB model has four more formed backbone hydrogen bonds but two less formed axial charge pairs, as shown in Figure 3 and Table 2. This again is surprising because the more disordered helix, according to the relaxation values, has more backbone hydrogen bonds formed. However, looking at the helical twist of the mutated triple helices gives additional insight into this issue. The helical twist is the angle (in degrees) that the triple helix rotates from one cross section to the next (see the Supporting Information and Experimental Section for a detailed description). For a general (POG)10 triple helix, the helical twist varies between approximately 95 and 115° from one cross-section to the next. In the B-Ala case (Figure 5), the helical twist varies more dramatically with values from 70 to 130°, with many swings from smaller to much larger angles. This shows a helix with a tightly twisted cross sections (larger angle) adjacent to loosely twisted (smaller angle) cross sections. However, while the A-Ala case has variation, it is primarily to a larger angle near the mutation site. We propose that the tightly twisted region around the mutation site allows the triple helix to compensate for the steric effects of the mutation, while the B-Ala case is more constrained by formed axial charge pairs. These charge pairs force the variation in helical twist, resulting in a triple helix with more interactions but that is more constrained and relatively inflexible. These systems show the demands of charge-pair formation to the D

DOI: 10.1021/acs.biomac.6b01808 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 3. Interactions in (a,b) B-Ala,13 (c,d) A-Ala, and (e,f) 3-Ala. (a,c,e) Models of the mutation site. Yellow amino acids are the mutated alanines. The red chain is the leading strand, the green chain is the middle strand, and the blue chain is the lagging strand in the triple helix. Hydrogen bonds are shown in the yellow dashed lines. (b,d,f) Schemes of the interactions found in the models. Blue dotted lines are expected backbone hydrogen bonds that are formed in the final model. Green dotted lines are expected charge pairs that are formed in the final model. Red dotted lines show expected interactions that are not formed in the final model. Purple dotted lines are compensating interactions that appear in the final models that do not form in the unmutated model. Asterisks denote the isotopically labeled glycines. R1 denotes a lysine side chain, R2 denotes a glutamic acid side chain, and R3 denotes an aspartic acid side chain.

triple helix as well as the flexibility needed to compensate for a glycine mutation. A-Ser melted at 30 °C, similar to the B-Ser triple helix that melted at 29.5 °C, and B-Ala melted at 30.5 °C. The similar melting point is surprising because in the models B-Ala and BSer have formed seven and six highly stabilizing charge pairs, respectively, while B-Ser has formed only five. The relaxation values of the 1H,15N-HSQC peaks in the A-Ser spectrum were similar to the A-Ala cases, suggesting a moderately disordered helix. These relaxation values are lower than the B-Ala and BSer, suggesting a less ordered helix with a similar melting point. When comparing the A-Ser and A-Ala spectra, much less A-Ser was seen to fold by NMR with a relative volume of 0.18 trimer, suggesting that this mutation not only affects the structure and composition of the assembled peptide but also negatively influences its folding equilibria (Table 1). A-Ser was also modeled using constraints from NMR. The ASer A′BB helix was similar to the A-Ala A′BB helix with one lost hydrogen bond and five axial charge pairs formed (Table 2). However, A-Ser had additional hydrogen bonds formed that were not present in A-Ala, three of which were between the serine OH side chain and the peptide backbone (Figure 4).

These additional interactions stabilize the triple helix but also constrain the area around the mutation. This can be seen when comparing the A-Ser and A-Ala helical twist, where the unconstrained A-Ala has one increase to 130° and quickly returns to normal within three cross sections; however, the slightly more constrained A-Ser has a more diffuse tight twist region that extends to 125° but stays elevated and more variable for six cross sections. This more constrained area around the mutation would make it more inflexible, leading to the observed lower melting point of 29.5 °C compared with AAla at 34.0 °C. Triple Mutant Helices. Unlike A-Ala and A-Ser, 3-Ala and 3-Ser fold into the designed A′A′B′ composition and register triple helix. Both have four peaks of equal volume in the 1 15 H, N-HSQC, indicating four distinct isotopic labels in the trimer (labeled G1−G4), as expected in the A′A′B′ triple helix and analogous to the base system. These systems, however, are significantly less stable and less ordered than the base triple helix, having melting points of 19 and 18 °C, respectively. These are lower than either A-Ala or A-Ser and are the lowest melting point of any of the peptides investigated (Table 1, Figure S11). E

DOI: 10.1021/acs.biomac.6b01808 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 4. Interactions in (a,b) B-Ser,13 (c,d) A-Ser, and (e,f) 3-Ser. (a,c,e) Models of the mutation site. Yellow amino acids are the mutated serines. The red chain is the leading strand, the green chain is the middle strand, and the blue chain is the lagging strand in the triple helix. Hydrogen bonds are shown in the yellow dashed lines. (b,d,f) Plots of the interactions found in the models. Blue dotted lines are expected backbone hydrogen bonds that are formed in the final model. Green dotted lines are expected charge pairs that are formed in the final model. Red dotted lines show expected interactions that are not formed in the final model. Purple dotted lines are compensating interactions that appear in the final models that do not form in the unmutated model. Asterisks denote the isotopically labeled glycines. R1 denotes a lysine side chain, R2 denotes a glutamic acid side chain, and R3 denotes an aspartic acid side chain.

3-Ala has relaxation values (5.3, 2.4, 3.0, and 4.5) below that of A-Ala (7.7, 8.9, 11.9, 7.8, and 6.0) and substantially lower than the base triple helix, indicating a very disordered region around the mutation. Additionally, there is significantly less helical population compared with the monomer (0.19 vs 0.81), lower even than A-Ala (Table 1). 3-Ala also appears to adopt a second minor register, as evidenced by the presence of an additional set of triple helix peaks in the 1H,15N-HSQC (Figure 2). There are a number of possibilities for the alternate register, but because of low folded populations and overlapping chemical shifts, further characterization was not possible. The model of the 3-Ala A′A′B′ helix was surprisingly wellfolded, considering that this helix must incorporate three mutations, with only one lost hydrogen bond and nine formed charge pairs. The lost hydrogen bonds and charge pairs are in the mutated region, which suggests a local disruption that compensates for the mutations. The helical twist of A-Ala agrees with this hypothesis, with a shift toward larger twist angles, up to 130° around the mutated region extending for five cross sections. This suggests a perturbed and inflexible helix that has difficulty compensating for the three mutations, leading to the low melting point.

3-Ser also retained composition and register control to form the A′A′B′ triple helix and also showed a drop in relaxation values (8.3, 4.8, 5.3, and 5.8) when compared with A-Ser, indicating greater conformational freedom (Table 1). 3-Ser had a fraction of trimer to monomer of 0.17 which is similar to ASer. This is notable because 3-Ser has three mutated amino acids and is still able to fold a comparable number of triple helices as A-Ser, which only has one mutation. In the 3-Ser A′A′B′ model, two hydrogen bonds and one charge pair are lost, similar to the 3-Ala model. However, the serine OH side chains contribute an additional four hydrogen bonds between the adjacent OH groups and with the peptide backbone leading to a more constrained mutated region. The helical twist data also suggest a more constrained mutated region with values extending over 130° and under 80°, with the disruption extending over five cross sections. This suggests a highly perturbed yet inflexible structure, which is compatible with the low melting point.



CONCLUSIONS The designed triple helical systems presented in this paper show that additional mutations in the constituent peptide F

DOI: 10.1021/acs.biomac.6b01808 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules

Figure 5. Helical twist of the peptide systems. The first (leading) chain of the triple helix is in red, the second (middle) is in green, and the third (lagging) is in blue. The vertical gray area is the mutated region, and the horizontal gray area is the range of helical twist in a POG model. The x axis is the cross-section of the triple helix, starting where the lagging strand joins the triple helix.



strands of a triple helix result in a drop in melting temperature. In all cases the triple-helix folding is affected by two major factors: (1) the number of mutations in the triple helix and (2) the stability and inflexibility conferred by axial charge pairs. The battle between these factors is clear in the A-Ala and A-Ser cases, where the designed A′A′B triple helix would have ten axial charge pairs and two mutations and nicely satisfy the 2:1 mixing ratio; the triple helix that actually folds, however, is the A′BB triple helix, with only five satisfied charge pairs but also only one mutation and still a significant portion of unfolded single chains. The number of mutations present clearly outweighs the stabilization from charge pairs; therefore, the mutation directs the composition of the triple helix. In contrast, the 3-Ala and 3-Ser systems only have mutated peptides available, which forces the formation of a triple helix with three mutations, and the one with the most axial charge pairs, A′A′B′, forms. There is no alternative system where fewer mutations could be incorporated. Despite retention of the designed composition and register, these are weakly folded systems with the largest amount of unfolded peptide. In the folded systems, the charge pairs are in direct opposition to the flexibility needed to incorporate the mutations. This tension leads to wild variation in the helical twist and an overall inflexible and unstable region. However, when fewer formed charge pairs are present, such as the comparison between B-Ala and A-Ala with seven and five formed charge pairs, respectively, some of the tension on the helix is lifted, allowing for a locally disrupted helix, which can normalize over fewer cross sections than the more constrained B-Ala helix.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.biomac.6b01808. List of constraints used for modeling, mass spectra, HPLC chromatograms, circular dichroism spectrum, additional NMR spectra, and a discussion of the assignment process. (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Jeffrey D. Hartgerink: 0000-0002-3186-5395 Author Contributions †

K.A.C. and A.M.A.-J. contributed equally.

Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported in part by the Robert A. Welch Foundation Grant C1557 and the National Science Foundation DMR-1206899.



REFERENCES

(1) Ramachandran, G. N.; Sasisekharan, V. Nature 1961, 190, 1004− 1005. (2) Rich, A.; Crick, F. H. C. J. Mol. Biol. 1961, 3, 483−506. G

DOI: 10.1021/acs.biomac.6b01808 Biomacromolecules XXXX, XXX, XXX−XXX

Article

Biomacromolecules (3) Kielty, C. M.; Hopkinson, I. In Connective Tissue and Its Heritable Disorders, 2nd ed.; Royce, P. M., Steinmann, B., Eds.; John Wiley & Sons Inc.: Hoboken, NJ, 2003; p 159. (4) Marini, J. C.; Forlino, A.; Cabral, W. A.; Barnes, A. M.; San Antonio, J. D.; Milgrom, S.; Hyland, J. C.; Korkko, J.; Prockop, D. J.; De Paepe, A.; Coucke, P.; Symoens, S.; Glorieux, F. H.; Roughley, P. J.; Lund, A. M.; Kuurila-Svahn, K.; Hartikka, H.; Cohn, D. H.; Krakow, D.; Mottes, M.; Schwarze, U.; Chen, D.; Yang, K.; Kuslich, C.; Troendle, J.; Dalgleish, R.; Byers, P. H. Hum. Mutat. 2007, 28, 209− 221. (5) Byers, P. H.; Wallis, G. A.; Willing, M. C. J. Med. Genet. 1991, 28, 433−442. (6) Beck, K.; Chan, V. C.; Shenoy, N.; Kirkpatrick, A.; Ramshaw, J. A. M.; Brodsky, B. Proc. Natl. Acad. Sci. U. S. A. 2000, 97, 4273−4278. (7) Radmer, R. J.; Klein, T. E. Biochemistry 2004, 43, 5314−5323. (8) Brodsky, B.; Thiagarajan, G.; Madhan, B.; Kar, K. Biopolymers 2008, 89, 345−353. (9) Bella, J.; Eaton, M.; Brodsky, B.; Berman, H. M. Science 1994, 266, 75−81. (10) Bella, J.; Brodsky, B.; Berman, H. M. Structure 1995, 3, 893−90. (11) Li, Y.; Brodsky, B.; Baum, J. J. Biol. Chem. 2009, 284, 20660− 20667. (12) Gauba, V.; Hartgerink, J. D. J. Am. Chem. Soc. 2008, 130, 7509− 7515. (13) Acevedo-Jake, A. M.; Clements, K. A.; Hartgerink, J. D. Biomacromolecules 2016, 17, 914−921. (14) Jalan, A. A.; Hartgerink, J. D. Biomacromolecules 2013, 14, 179− 185. (15) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.; Bax, A. J. Biomol. NMR 1995, 6, 277−293. (16) Vranken, W. F.; Boucher, W.; Stevens, T. J.; Fogh, R. H.; Pajon, A.; Llinas, M.; Ulrich, E. L.; Markley, J. L.; Ionides, J.; Laue, E. D. Proteins: Struct., Funct., Genet. 2005, 59, 687−696. (17) Rohl, C. A.; Strauss, C. E. M.; Misura, K. M. S.; Baker, D. Methods Enzymol. 2004, 383, 66−93. (18) Leaver-Fay, A.; O’Meara, M. J.; Tyka, M.; Jacak, R.; Song, Y.; Kellogg, E. H.; Thompson, J.; Davis, I. W.; Pache, R. A.; Lyskov, S.; Gray, J. J.; Kortemme, T.; Richardson, J. S.; Havranek, J. J.; Snoeyink, J.; Baker, D.; Kuhlman, B. Methods Enzymol. 2013, 523, 109−143. (19) Wolfram Research, Inc. Mathematica, 11th ed.; Champaign, IL, 2016. (20) Kabsch, W. Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1976, 32, 922−923. (21) Bella, J. J. Struct. Biol. 2010, 170, 377−391.

H

DOI: 10.1021/acs.biomac.6b01808 Biomacromolecules XXXX, XXX, XXX−XXX